REVIEW OF USPTO HEARINGS IN WASHINGTON ON SOFTWARE PATENTING Feb 12, 1994 Gregory Aharonian Internet Patent New Service P.O. Box 404, Belmont, MA 02178 617-489-3727 patents@world.std.com Last week the Patent and Trademark Office held two days of hearings at the Crystal Forum in Crystal City, in which 28 testified for approximately 11 minutes each during the two days. Commissioner Lehman, along with about a dozen other PTO officials, including Gerald Goldberg, director of Group 230 (which examines most of the software patents) were in attendance. About 70 people attended the first day, and 30 people the second day. (I should note that the second day, ALL of Washington, DC, was shut down because of a snow storm. Despite the shutdown of the federal government, the Commissioner and some of his colleagues braved the weather and traveled to the conference hall to hold the second day of hearings for those of us still in attendance [many trapped because of cancelled airplanes]. Would that all of the government was so dedicated.) The testimony was quite interesting, which I will summarize briefly before presenting summaries of each speaker's comments. I will also offer predictions to what will happen to software patenting activities. (Actually, my comments and predictions from the San Jose hearings review that I posted pretty much remain the same. Much of the commentary from the DC hearings reflected much of what was said at San Jose. If you missed my review of the San Jose with my observations, let me know and I will email a copy to you.). ============================================================================== SUMMARY COMMENTS OF TESTIFIERS AT THE HEARINGS Thursday, February 10 (in order of testimony) Paul Robinson, Programmer, Tansin Darcos & Co. I collect software to reuse and learn. Companies wage vicious software marketing wars. New releases have to be rapid to stay in business [with the implication that that success is more important than patent protection]. Software represents mental processes, cost free electron manipulation in microprocessor circuits (with no manufacturing/costs needed). Systems are large and costly to modify, especially when forced by infringement. Against first-to-file. Hard to check existing and new software patents. Institute compulsory license. Eliminate secrecy - allow pre-publication. Keith Stephens, Corporate Counsel, Taligent Have to transform legal chaos of software patents - eliminate uncertainty. Hire computer science majors with industry experience as examiners. Applaud training efforts for 2300 group. Create human database of experts for software prior art issues. Give examiners better tools like Internet access. Tune examination process. Allow patents to be filed electronically. Modify reexamination process - make more open, less biased towards patent holder. Pre-publication is not necessary. Mark Traphagen, Counsel, Software Publishers Association SPA to date primarily focused on copyright law. However limited only to expression, not enough for software. SPA has 1100 members. Improve patent examination process first, then try changing statutory laws. Will ask members to donate software prior art to the Patent Office, and to help train examiners. Expanded prior art collections needed. Will award scholarships to examiners to attend software conferences to interact with industry. Rob Lippincott, Interactive Multimedia Association IMA has 200 multimedia publishers. Interactivity not a patentable process. They are trying to reduce barriers to multimedia use. Members fear patents which constrain flow of information. Patent system indifferent to impact on industries. Broad patents should not be granted, they are inherently regulatory in nature. Prefer pre-publication. Need better prior art databases. Average cost of litigation is $500,000 per claim. Insurance costs will rise and be prohibitive. [As a suggestion to the SPA, IMA and LPF, you can strengthen your arguments much more if you did a little a historical research, and referenced the RCA and Hazeltine patent pools, an approach the MPEG people are now considering.] E. Robert Yoches, Finnegan Henderson Farabow Garrett & Dunner, Attorney Software patent examination should be same as for other industries. Wouldn't be practical to do so, encourages lawyers to play games with claims drafting. Extra attention would complicate litigation by creating double standards. Nothing fundamentally wrong with software patents. Need better prior art, and retain qualified examiners. Shouldn't ask applicants for more background information. Standards of novelty and obviousness should match industry standards in software, but applied with the same rigor as in other technology areas. Stephen Noe, Intellectual Property Owners Inc. IPO supports software patents. Software is not different from other technologies in terms of patenting. Hard to define what software is for use in legal considerations. Continued patent protection for software important to US economy. No need for special obviousness and novelty tests. Need better prior art. Train examiners in software technologies. Special duty for disclosure for software applicants not fair or practical, would encourage people to claim ideas as non-software. Use the Internet. Industry needs "insurance" provided by patent protection. Software is not a different type of technology. John Horn, Allen-Bradley, Patent Counsel Industrial control hardware and software have similar claims. Thus software should be patentable, but most look at relevant hardware prior art when considering software patent applications. Need better prior art. Software does not need higher standards of reliability. Richard Nydegger, Workman Nydegger & Jensen, Attorney Samual Morse's and Alexander Graham Bell's patents' broad claims were upheld in court, leading to new industries despite sweeping nature of claims. Broad protection is justifiable. Current system for software patents does have problems, which can be fixed. Improve access to prior art. Better classify software technology (such as systems used by IEEE and ACM). Train examiners better, who need more exposure to software technology. Allow third party involvement in reexamination. PTO should decrease pendency time for patent applications. After 18 months allow pre-publication - helps prevent "submarine" patents. Allan Ratner, Philadelphia Patent Law Association, President & Attorney Hard to define what software industry, and software itself, is. Growing use of software and embedded microprocessor chips helps blur distinctions. Impossible to separate software from its embedded hosts. Copyright not sufficient protection. Higher patenting standards for hardware/software patents unfair and impractical. Examiners need better prior art. Pre-publication one way to solicit third-party prior art. Open up the reexamination process. Diane Callan, Lotus Development Corporation, Deputy General Counsel Business Software Alliance Software has rapid innovation and widespread use. Software industry is a major contributor to US economy. Strong IP protection for software is needed. Source code vital to companies needing protection. No new forms of protection desired for software. Novelty and obviousness standards must be more rigorously enforced. Patent pendency too long, partially due to continuations. Protection should be fixed-term from filing date. Better prior art database is needed. Give examiners better searching tools. Should require applicants to do comprehensive prior art search. Allow third party into reexamination - allow opposition period. R. Duff Thompson, WordPerfect Corporation, General Counsel Software piracy endemic. Intellectual property protection needed. Changing laws will take to long to pass Congress. Existing laws provide an adequate framework. Need better prior art database, and faster patent application processing. WordPerfect pledges to help patent office. Supports having an opposition period. Endless continuations delay patents while other companies start using similar technologies, technologies then incoporated into revised patent claims and greatly expanding scope of industry infringement. Prefer fixed term for patent protection from filing date. Supports pre-publication. Don't like Doctrine of Equivalents greatly expanding scope of software claims. WordPerfect now filing defensive software patents. Richard Jordan, Thinking Machines, Patent Counsel Thinking Machines' Connection Machines intimately relies on both hardware and software. 70% of their R&D staff are software developers, who are encouraged to publish papers without prior legal review. Patent protection important for protecting its investments. Copyright insufficent for protecting its investments and preventing reverse engineering. However, the software patenting process can be made better. Patent prosecuting pendency too long - would like pre-publication. Hard to get prior art from foreign patent offices. Opposition period would allow public to be involved. Jeff Berkowitz, Finnegan Henderson Farabow Garrett & Dunner, Attorney PTO should not force people to submit source code listings. Ron Reiling, Digital Equipment Corporation, Counsel Patent protection for software important. Patents allow ideas to be assets, essential for information age. Software inventions should be treated the same as other technologies. Better training is needed for examiners. All software concepts should be patentable. Standards for software patents should be same as other technologies. Software ideas, though, must be novel and non-obvious. No special rule-56 patentability searches should be imposed on software patent applicants. Program code listings should be accepted. PTO invest in having quality examinations. Need more examiners. Have better explanations of office-action claims rejections. Need better software prior art database. Open up reexamination process. DEC willing to work with PTO to help improve processes. A. Jason Mirabito, Boston Patent Law Association, Board Member Needs more funds for training examiners and better prior art, but don't raise fees [note: for those who don't know, the PTO is one of the few government agencies that funds its budget from user fees, no Congressional appropriation]. Hire more computer scientists as examiners, as did the biotech group. Open reexamination to third parties. Adopt pre-publication system, but after first office action (which sometimes is later than 18 months). No special disclosure standards for software. No special prior art search burdens for software applicants. Section 112 best guidelines for describing software inventions. Jonathan Band, Morrison & Moerster, Attorney Better to treat "bad software patents" than "software patents are bad". Consider pre-publication and fixed-term patent rights after filing. Patent lawyers benefit too much from current chaos. Bring in computer experts. However, overprotecting software intellectual property as bas as underprotecting software intellectual property. Leonard Suchtya, Bellcore, General Attorney Bellcore uses patents, trade secrets and copyrights for software, so strongly favors patent protection for software inventions. Patents and printed publications provide plentiful prior art. Bellcore will donate their publications to the PTO library. Standards can be established for novelty and obviousness. Rule 56 requirements sufficient, no special need for different rules for software. No need for applicants to perform special prior art search. Filing source code shouldn't be necessary, under section 112, and sometimes for large lines-of-code programs hides interesting features. In fact, filing source code should not be permitted. Vern Blanchard, Software Entrepreneur, San Diego Against software patents, they must be eliminated. Programmers don't like litigation, would rather switch fields. New companies could be barred without lots of licenses. His company developed a Bingo program, working for two years without pay, and whose customers liked it a lot. Then a competitor sued over patent infringement, and he was put out of business. Competitor claims covered handheld device, and they filed for injunction. Blanchard is now in debt for $100,000, though injunction was overturned. Competitor spent $450,000. In the end, the legal fees killed his company, even though they were in the right. Did consider reexamination, but probably would have had little effect on their court case. Patent run counter to spirit of sharing of software over the Internet. Ed Currie, Imagesoft, Software Entrepreneur Involved with patent infringement suit. Need better prior art database. His company has spent $120,000 just arguing over venue for lawsuit. Want to use reexamination, but 6 law firms in New York advised against reexamination, saying it favors patent holder and hinders further prosecution abilities. Friday, February 11 (in order of testimony) D.C. Toedt, Arnold White & Durkee, Attorney PTO should try using some of the disclosure practices of the SEC, where prior corporate financial information is similar to prior invention info. Patent examinations should be more timely, cutoffs for discovery, and fixed dates to resolve issues. Christian Hofstader, League for Programming Freedom, President LPF opposed to software patents. As size of industry grows, length of patent life should drop. They believe "software" can be defined (as opposed to some at San Jose [including myself] that argued that hardware and software are growingly indistinct), so that legislation can be drafted banning software patents. Microcomputer software leaders grew industry without using patents, why do we need them? Copyright is sufficient for software intellectual property protection. Copyright allows companies to compete via rival implementations, which drives innovation. Tim Scanlon, Allen Bradley (AB) Hardware functionality being replaced by software functionality. If one form deserves patent protection, so then does the other. GUI technology is important to AB. Icon, bitmaps and control box components important to AB products, especially multilanguage interfaces and symbology for global recognition. A lot of ergonomic research goes into their products to make them more usable. Copyright protection not sufficient, concentrates too heavily on expression. Strengthen patent protection for GUI componentry. R. Lewis Gable, Welsh & Katz, Patent Attorney Experience level of average software examiner is low. Fewer primary examiners at PTO, 1:10 to non-primary examiners, who mostly are inexperienced. Examiners don't have enough time to process applications and do prior art search. Younger examiners tend to cite prior art patents with little relevance to application under examination. In group 2300, 130 out of 160 are non-primary examiners, with 89 having less than 2 years experience. Examiners not paid enough, and after a few years experience, law firms offer bigger bucks and out they go. John DeWald, Prudential Insurance, General Counsel Methods of doing business, not patentable, is patentable when implemented in software. Fighting such lawsuits is costly and uncertain, costing millions in the financial community. This has chilling effect on customers. Patent holder, with biases in his favor, can wreck havoc. Examiners need better prior art, which should include documents of business practices. PTO should impose special duty on software patent applicants to disclose prior art. Computer implementations of existing business practices should not be patentable. Reexamination should be opened. Have an opposition period. David Clark, local patent attorney Formerly as examiner in group 2300 for ten years, now a private attorney. Key problem is retaining examiners for more than a few years. Provide better tools for examination process. Group 2300 most compete with private sector for hiring. Examiners don't have enough time. Current searching out of control. Better classify information for retrieval and dynamically altering classification scheme. Should use computer networks to exchange information, both internal to PTO and externally. Salaries can double for examiners if they goto private sector, a strong lure. Samuel Oddi, Professor Economic impact of intellectual property. Published an article on revolutionary patents and their economic impact. A patent free society would see lower rates of innovation and commercial exploitation. Revolutionary patents treated poorly by current patenting system - ideas are too new to satisfy statutory regulations. Copyrights are difficult to apply to software and revolutionary ideas, where expression far less important than the functionality. Risk taking business aspects of inventing require patent protection, including software. Bernard Galler, Professor Lots of software prior art not published in readily available sources. Hard to locate innovative and non-obvious source code prior art. Most development outside of academia. Commercial product manuals also prior art. Applauds efforts of Source Translation & Optimization to build the largest comprehensive software prior art database. =========================================================================== ( The last speaker of the hearing was myself, speaking on the complexities of software prior art database development and use. What follows are an extended version of my comments. Hey, its my newsletter :-) Greg Aharonian, Source Translation & Optimization, President There are six stages of the patenting process where knowledge of all existing publicly known information, prior art, is necessary for the patenting system to work as efficiently as possible: information disclosure document, patent examination, opposiiton period, reexaminations, patent infringement lawsuits, and circuit court decisions. At each of these stages, key issues are often obviousness and novelty, which can only be addressed by referencing a comprehensive prior art database. At these hearings, everyone has agreed that lack of such a database in the software field is causing many problems, including the issuance of bad software patents that lead to costly nuisance lawsuits and a dampening of innovation. The repeated calls by people testifying to address this issue, people who disagree on many other aspects of software patenting, illustrates how serious a problem the lack of a comprehensive software prior art database is. Some of the solutions proposed at the hearings, in particular using the Internet to solicit contributions of prior art for reexamination or opposition proceedings, without a database structure in place, will swamp the PTO with tons [sic] of documentation that people will have to sift through and bring order to, to relate it to the proceedings under consideration. For example, opening up all existing 11,000 software patents for reexamination, and asking for public contributions of prior art, would bring in hundreds of thousands of documents, many duplicative and lacking in prior art stature. A related problem in the software development community is the lack of a comprehensive reusable software components database, so that all of the benefits being promised of software reuse can be realized. While many tools have been developed to facilitiate software reuse, a key missing component has been large reusable component databases. Both problems, software prior art and reuse, have the same solution, a comprehensive of as much information as possible on software technology developed in this country over the last twenty five years. This country is now spending over $50 billion annually developing software (much by the US government), but spends very little keeping track of the results of these very large expenditures. And the tens of millions of dollars that has been spent trying to do so, for the most part, has achieved very little. In all cases, the reasons for such limited results are the same - people with little experience in building archival databases adopting passive approaches to collecting information, and underestimating the complexity of classification, maintenance and organization of tens and hundreds of thousands of documents that have to be dealt with. Too often, such efforts ignore the contributions of the library science community, who have decades of experience struggling with these issues. For the last eight years, I have building a software prior art/reuse database. Currently I have actively-collected information on over 15,000 computer programs available in source code form, 5000 software patents, and over 100,000 abstracts of entities of software technical information. It is the largest such database in the country, and has been developed privately by myself. One of the simpler questions is what to collect. Prior art appears in a variety of publications: government/academic/corporate technical reports, journal articles, conference proceedings, theses, books, commercial products through their manuals, Internet files, bulletin board files, and software patents. For each of these forms, there are legacies by the publishers that it is important to know, usually reflected in the editors and authors of items that appear in the publications. For example, I conclude different things knowing that an algorithm is published in IEEE transactions on software engineering versus being published in BYTE. In these various publication forms, there are a variety of styles in which prior art can occur: source code listings, object libraries, flowcharts, state charts, psuedo-code, mathematical algorithms, and patent claims. There are also some non-obvious forms of software prior art, such as VHDL and SPICE files (hardware description languages), and numerical data, which can be translated into algorithms. Thus there are over fifty combinations of publication forms and prior art styles that one has to be on the look for when gathering information for inclusion into a prior art database. Each combination has a different set of rules for determining whether or not to include an item in the database, and learning and applying these rules is not a trivial matter. These materials also have different physical forms. Printed materials, microfiche, microfilm, microform, magnetic tapes, CDROMs, diskettes, card decks, and lineprinter paper stacks all have to be dealt with, and when you are dealing with hundreds of thousands of items, you need large storage facilities. To both learn or apply such rules, you have to do something that has escaped most previous efforts to build these prior art/reuse databases. To really collect enough information to build a comprehensive database, you have to actively seek out the information, and not rely on passive donations of materials. While passive approaches to collecting information do result in some materials being collected, the vast majority of the information will not be obtained. In some cases, people and companies are lazy about contributing. In other cases, the information is not well organized at the hundreds of facilities around the country where software activities occur, so that even if someone wanted to voluntarily donate their materials, they would first have to invest a lot of time and effort in doing so. When I think of prior art, I think of dimly lit basements with shelves from the floor to the ceiling. The best way to collect information is to go after it. Over the past eight years, I have visited hundreds of government/academic/corporate sites, searching through their libraries looking for items useful for software prior art and reuse. One key problem in building and maintaining such databases is determining how to monitor so many sites cost effectively, especially given that some information is duplicated at many sites. You can waste lots of time reexamining materials you already examined somewhere else. Working out of the Boston and Silicon Valley communities (which are very rich in terms of sources of software information), I have been able to develop a regimen to monitor efficiently. It would be hard to do this outside of these areas, at least for locating software information. At these facilities, you have to one-by-one manually examine each item of potential prior art/reuse information, and when candidates are found, you then have to extract key information on its content for processing and inclusion into the database. This can be very mind numbing; think of looking at many hundreds of hundred page university theses over the course of an afternoon or a week of afternoons. I then have to drag interesting theses down a few floors to a copier to copy key sections to be summarized and categorized back at my office. Nothing else will work, despite attempts to disparage such efforts by calling them "software salvaging". It also needs to be done by experienced people, and obtaining experience in this matter is not easy. There are no courses for it, and no companies really doing it (as a place to learn). I learned empirically, by doing it for eight years. But without experience, it is impossible to prioritize where you search and what you search for and what you want to enter into a database. The only alternative is to throw hundreds of millions of dollars at the problem, which would work. For example, one area you have to acquire experience for, are schemes for classifying software knowledge. The Patent Office has one scheme, the Library of Congress another, the Dewey Decimal system for non-LC libraries, the IEEE has one, the ACM has a few, online databases have their own schemes, and then there are other schemes not used for software, but worth knowing, to help you think about these issues, including the Encyclopedia Brittanica classification, the work of James Grier Miller. There are also dozens of specialized schemes from smaller software/mathematical societies, and a variety of academic efforts to prepare taxonomies. Despite being in the era of gigabyte-a-second keyword text searching retrieval systems, the need for classification schemes is important. In short, building software prior art/reuse databases is neither trivial or something for amateurs. And as we speak, the complexity is becoming even greater, for two reasons. First the growing capabilities of hardware and software codesign, in which representation languages such as VHDL represent both hardware and software, make it possible for hardware to serve as prior art for software. Since there is at least twice as much hardware information out there as software information (in the formerly strict sense of their definitions), this convergence triples the size of a comprehensive database. Second, the NAFTA and GATT treaties will allow foreigners to have first-to-invent rights when filing in the United States (currently when they file US patents, they only have first-to-file rights). To settle infringement cases where foreigners are involved will then require a prior art database that encompasses the entire planet. Based on patent counts, this will also double the size of the database, as well as aggaravate the problems of cost effectively visiting hundreds of sites around the world searching for prior art. In short, the problem of building a software prior art database is very complex. The rewards to the US economy, however, are even larger, so that doing so is both necessary to treat the growing problems with software patents while being beneficial in the resulting large body of software information so collected. Driving the need to build the database will be the accelerating rate of both software patent applications and awarded patents, and patent infringement lawsuits. While I believe that on its merits, the Compton case is making a mountain out of a mole hill, cases like Microsoft/Stac and Novell/Billings will less and less be the exceptions.